1/n
I’m really excited to share that our @OpenAI reasoning system got a perfect score of 12/12 during the 2025 ICPC World Finals, the premier collegiate programming competition where top university teams from around the world solve complex algorithmic problems. This would have placed it first among all human participants. 🥇🥇
I’m really excited to share that our @OpenAI reasoning system got a perfect score of 12/12 during the 2025 ICPC World Finals, the premier collegiate programming competition where top university teams from around the world solve complex algorithmic problems. This would have placed it first among all human participants. 🥇🥇
2/n
We officially competed in the onsite AI track of the ICPC, with the same 5-hour time limit to solve all twelve problems, submitting to the ICPC World Finals Local Judge - judged identically and concurrently to the ICPC World Championship submissions. We received the problems in the exact same PDF form, and the reasoning system selected which answers to submit with no bespoke test-time harness whatsoever. For 11 of the 12 problems, the system’s first answer was correct. For the hardest problem, it succeeded on the 9th submission. Notably, the best human team achieved 11/12.
We officially competed in the onsite AI track of the ICPC, with the same 5-hour time limit to solve all twelve problems, submitting to the ICPC World Finals Local Judge - judged identically and concurrently to the ICPC World Championship submissions. We received the problems in the exact same PDF form, and the reasoning system selected which answers to submit with no bespoke test-time harness whatsoever. For 11 of the 12 problems, the system’s first answer was correct. For the hardest problem, it succeeded on the 9th submission. Notably, the best human team achieved 11/12.
3/n
We competed with an ensemble of general-purpose reasoning models; we did not train any model specifically for the ICPC. We had both GPT-5 and an experimental reasoning model generating solutions, and the experimental reasoning model selecting which solutions to submit. GPT-5 answered 11 correctly, and the last (and most difficult problem) was solved by the experimental reasoning model.
We competed with an ensemble of general-purpose reasoning models; we did not train any model specifically for the ICPC. We had both GPT-5 and an experimental reasoning model generating solutions, and the experimental reasoning model selecting which solutions to submit. GPT-5 answered 11 correctly, and the last (and most difficult problem) was solved by the experimental reasoning model.
4/n
This result is a great capstone to our streak of results showcasing the impressive pace of improvement of our reasoning systems. The same set of models also competed in the IMO and IOI, highlighting the generality of these systems.
This result is a great capstone to our streak of results showcasing the impressive pace of improvement of our reasoning systems. The same set of models also competed in the IMO and IOI, highlighting the generality of these systems.
5/n
This is just the beginning, and I’m personally very excited about our research roadmap. The next frontier is the discovery of new knowledge, which is the true milestone at the end of the day.
This is just the beginning, and I’m personally very excited about our research roadmap. The next frontier is the discovery of new knowledge, which is the true milestone at the end of the day.
6/n
The competition itself was electric, a confluence of many of the smartest young minds in the world. Huge congrats to all of this year’s participants! Thanks to the ICPC committee for organizing such a great competition and for allowing us to compete. And kudos to the ICPC judges for constructing such intellectually challenging, clear, and well-written problems!
The competition itself was electric, a confluence of many of the smartest young minds in the world. Huge congrats to all of this year’s participants! Thanks to the ICPC committee for organizing such a great competition and for allowing us to compete. And kudos to the ICPC judges for constructing such intellectually challenging, clear, and well-written problems!
7/n
The city of Baku is beautiful, an ancient city at the crossroads of many historic civilizations. We really enjoyed exploring the city in our spare time. Here we are touristing at the Palace of the Shirvanshahs 😀
The city of Baku is beautiful, an ancient city at the crossroads of many historic civilizations. We really enjoyed exploring the city in our spare time. Here we are touristing at the Palace of the Shirvanshahs 😀
8/n
Research is very much a team sport, and it was so much fun to work with @wyndwarrior, @robin_a_brown, and @bminaiev on prepping for the competition—and with @_lorenzkuhn, @clavera_i, @andresnds, and @ahelkky on competitive programming in general. I remember when I joined OpenAI, our best AI models could barely solve Codeforces easy problems. It’s incredible to see the rate of progress over these last two years. And with this team, I’m really excited to see what we achieve in the next two!
Research is very much a team sport, and it was so much fun to work with @wyndwarrior, @robin_a_brown, and @bminaiev on prepping for the competition—and with @_lorenzkuhn, @clavera_i, @andresnds, and @ahelkky on competitive programming in general. I remember when I joined OpenAI, our best AI models could barely solve Codeforces easy problems. It’s incredible to see the rate of progress over these last two years. And with this team, I’m really excited to see what we achieve in the next two!
• • •
Missing some Tweet in this thread? You can try to
force a refresh
